Data-driven Adaptive Robust Optimization for Resource Sharing During a Pandemic

Abstract

Outbreaks of COVID-19 in local communities yield a massive increase in demand for limited resources such as ICU beds and mechanical ventilators. To cope with such outbreaks, many hospitals canceled or postponed elective procedures to preserve care capacity (including life-saving ventilators) for patients with COVID-19. This resulted in a substantial financial loss for the hospitals and poor outcomes for non-COVID-19 patients. It is worth noting that the infection spreads at varying rates in different states and counties. This provides an opportunity for sharing scarce resources such as ventilators, which can be transported over large distances, among states and counties to alleviate capacity shortfalls caused by an epidemic surge in a state or county. In this paper, we develop a novel data-driven adaptive robust optimization methodology for the allocation and relocation of mechanical ventilators among different states and counties. This methodology considers uncertainty in the rate of disease spread, and therefore demand for ventilators, in various states and counties by leveraging a powerful microsimulation model of COVID-19 spread and intervention. Our main theoretical contribution lies in a new policy-guided model, which mitigates some critical limitations of current robust policies and stochastic programming approaches and makes the resource sharing decisions implementable in practice. Proof of concept will be given for the allocation and relocation of ventilators among a subset of states in the U.S. Our method can be adapted to any state or county and can consider other portable resources such as healthcare personnel, personal protective equipment, and point-of-care testing units.

Keywords: COVID-19, Pandemic, Resource Sharing, Elective Surgeries, Data-driven Adaptive Robust Optimization, Microsimulation.